Spectrin-dependent and -independent Association of F-Actin with the Erythrocyte Membrane

Transcription

1 Published Online: 1 August, 1980 Supp Info: Downloaded from jcbrupressorg on November 11, 2018 Spectrin-dependent and -independent Association of F-Actin with the Erythrocyte Membrane CARL M COHEN and SUSAN F FOLEY Department of Research, Division of Hematology/Oncology, St Elizabeth's Hospital, Boston, Massachusetts and Department of Medicine, Tufts Medical School, Boston, Massachusetts ABSTRACT Binding of F-actin to spectrin-actin-depleted erythrocyte membrane inside-out vesicles was measured using [3H]F-actin F-actin binding to vesicles at 25 C was stimulated 5-10-fold by addition of spectrin dimers or tetramers to vesicles Spectrin tetramer was twice as effective as dimer in stimulating actin binding, but neither tetramer nor dimer stimulated binding at 4 C The addition of purified erythrocyte membrane protein band 4 1 to spectrin-reconstituted vesicles doubled their actin-binding capacity Trypsinization of unreconstituted vesicles that contain <10% of the spectrin but nearly all of the band 4 1, relative to ghosts, decreased their F-actin-binding capacity by 70% Whereas little or none of the residual spectrin was affected by trypsinization, band 4 1 was significantly degraded Our results show that spectrin can anchor actin filaments to the cytoplasmic surface of erythrocyte membranes and suggest that band 4 1 may be importantly involved in the association The role of intracellular actin filaments in the generation of cell motility and the maintenance of cell shape is well recognized, and there is considerable morphological and some biochemical evidence suggesting that microfilaments may function, at least in part, through their association with cell membranes (1, 13, 14, 15) Human erythrocytes, like other cells, contain membrane-bound actin as a major component of their cytoskeleton (17, 24) and, although membrane-bound actin filaments have not been detected morphologically, indirect evidence suggests that at least some of the actin is in the form of short filaments (4, 8, 18) An important role for these actin filaments in the erythrocyte cytoskeleton was suggested by experiments in which tetramers of spectrin, the major erythrocyte cytoskeletal protein, with or without erythrocyte band 4 1, cross-linked actin filaments and formed three-dimensional viscous gels or oligomeric complexes (see footnote t and references 5, 9, 11 and 28) These experiments showed that F- actin can bind to spectrin and provided some insight into the assembly of the erythrocyte cytoskeleton It remains unclear, however, whether erythrocyte actin is bound to the membrane only through its association with spectrin or whether independent associations can be formed with other membrane components Here, we show that spectrin tetramers or dimers can stimulate binding of preformed F-actin to erythrocyte insideout vesicles, and that F-actin can also bind to spectrin-depleted membrane vesicles via a trypsin-sensitive interaction ' Cohen, C M, and C Korsgren Manuscript submitted for publication 694 MATERIALS AND METHODS Preparation of Actin Actin was prepared from rabbit skeletal muscle by the method of Spudich and Watt (20) and was used within 5 d after extraction from acetone powder [''H]actin was prepared by the method of Cohen et al (7), except that G-actin was chromatographed on Bio-Gel P-4 (Bio-Rad Laboratories, Richmond, Calif) using 2 mm N-tris(hydroxymethyl)methyl-2-aminoethane sulfonic acid (TES), ph 76, 05 mm ATP, 01 mm CaCI_, as the eluting buffer to remove traces of,8-mercaptoethanol before labeling with [''HIn-ethylmaleimide The specific activity of the actin ranged from 800 to 2,100 cpm/lag Actin was polymerized at a concentration of 5 0 mg/ml before binding experiments by the addition of 100 mm KCI and 2 mm MgCl, (final concentrations) and incubation at 25 C for 30 min Preparation of Spectrin Spectrin dimers and tetramers were prepared by minor modifications of established methods (5, 16, 19, 27) Erythrocyte ghosts prepared as described by Steck and Kant (23) were washed once in 30 vol ofice-cold 01 mm NaPO a, ph 7 6 The pelleted ghosts were resuspended with 0 5 vol of the same buffer and incubated at 37 C for 40 min to extract spectrin dimers and then sedimented at 225,000 g for 30 min The supemate was chromatographed at 4 C in a 2 5 x 100- cm column of Sepharose 4B using 0 l M NaCl, 25 mm Tris-HCI, ph 76, and 01 mm EDTA as the eluting buffer The dimer fractions (Fig 2, Panel I) were pooled, concentrated by ultrafiltration to -A_ = 1 0, and either used directly after 18-h dialysis against 20 mm KCI, 2 mm TES, ph 76 or incubated at 29 5 C for 2 h to promote conversion of dimer to tetramer (27) The incubated samples were then rechromatographed as described above, and the fractions containing spectrin tetramer (Fig 2, Panel II) were pooled, concentrated, and dialyzed as described above This two-step procedure for the preparation of tetramers was adopted to minimize potential contamination by void volume fractions which frequently spillover into tetramer fractions when tetramers are prepared by direct chromatography of 4 C membrane extracts (5, 27) THE JOURNAL Of Celt Biotocy " VOLUME 86 AUGUST The Rockefeller University Press /80/08/0694/05 $1 00

2 Preparation of Band 4 1 Band 4 1 was extracted from spectrin-depleted vesicles by minormodification of established methods (25, 28) Membranesremaining afterextraction of spectrin by dialysis at 4 C were washed once in 30 vol of 150 mm NaCl, 5 mm NaP0,, ph 76, and resuspended to 1 5 times their packed volume in this buffer The membranes were then extracted by the addition of 0 8 M KCI, 01 mm dithiothreitol (DTT), and 3 mm phenylmethyl sulfonyl fluoride (PMSF; final concentrations) and incubated at 4 C for 6-18 h with stirring After extraction, the membranes were sedimented at 225,000 g for 30 min and the supernate was desalted on a 3 x 10-cm Bio-Gel P4 column equilibrated with 5 mm Napo,, ph 76, 1 mm EDTA, 20 mm KCI, and 05 mm DTT The void volume fractions were then loaded into a 1 x 5-cm column of DEAE cellulose that had been preequilibrated with the buffer described above, and the column was flushed with 10 ml ofthe same buffer The column was then flushed with ml of 5 mm NaP0,, ph 76, 1 mm EDTA, 65 mm KCI, and 05 mm DTT and finally with 5 mm NaPO,, ph 76, 1 mm EDTA, 120 mm KCI, and 0 5 mm DTT, which elutes band 41 Elution was monitored by the A ofthe column effluent and the fractions containing band 41 were concentrated by ultrafiltration and dialyzed for 18 h against 20 mm KCI, 2 mm TES, ph 76, 0 1 mm EDTA, and 0 1 mm PMSF for use in binding experiments Typically, ml of protein at a concentration of 01 mg/ml were obtained from 50 ml of ghosts (see Fig 4 for gel of purified band 41) Preparation of Membranes Erythrocyte inside-out vesicles were prepared from ghosts (23) by incubation in 30 vol of03 mm NaP0,, ph 76, or 01 mm EDTA, ph 85, at 37 C for 30 min This simple treatment results in the elution of >90% of spectrin and actin and leads to formation ofsmall (<05-pin) membrane vesicles that morphologically are inside-out (23) These spectrin-actin-depleted inside-out vesicles are referred to in the text simply as vesicles Sealed ghosts were prepared by incubating freshly made ghosts in 5 mm NaPO,, ph 76, and 2 mm MgCl2 at 37'C for I h Measurement of Actin Binding Binding of F-actin was done by incubating membranes at protein concentrations ranging from 015 to 0 5 mg/ml in 50 mm KCI, 2 MM MgC12 05 MM ATP, 5 0 mm NaP0, ph 65, and 075 mm ß-mercaptoethanol (binding buffer) with the appropriate concentrations of actin at 25 C for 1 h After the incubation, triplicate 02-ml aliquots of the mixture were centrifuged at 12,500 rpm for exactly 10 min in a Sorvall SS34 rotor at 4 C (DuPont Instruments-Sorvall, DuPont, Co, Newtown, Conn) 005-ml aliquots of each supernate were counted for 'H, and the amount of ['HIF-actin in the pellet was computed from the difference between supernateand initial incubation mixture Standard deviations were generally 12-17% of the values shown in the figures Control experiments with vesicles labeled by phosphorylation with [y-'"p]atp by standard methods (2) showed that % of vesicles and ghosts sedimented under these conditions, whereas <10% of the actin found in the pellet with spectrin-reconstituted vesicles (see Results) sedimented in their absence The amount of actin sedimenting in the absence of vesicles was measured at each actin concentration in all experiments and was subtracted from the appropriate sample values Vesicles were reconstituted with spectrin by incubating 06 mg/ml vesicles at 4 C for I h in binding buffer (ph 76) containing the finalconcentrations of spectrin indicated in the figures, sedimented, and, after all supernate had been carefully removed, resuspended to their original volume for actin binding measurements Other procedures Gel electrophoresis in 5% SDS polyacrylamide was performed according to the method offairbanks et al (10), and protein bandswere labeled in accordance with standard nomenclature (10, 22) RESULTS Fig 1 shows that inside-out vesicles (Fig 1, gel a), which have 5-10% of the spectrin content of ghosts (Fig 1, gel c), bind added F-actin, and that reconstituting these vesicles with spectrin dimers (Fig 1, gel b) greatly increases their capacity to bind F-actin No binding of F-actin to sealed ghosts was detected, showing that, whether or not vesicles were reconstituted with spectrin, the binding was specific to the cytoplasmic membrane surface FIGURE 1 Binding of [ 3H]F-actin to sealed ghosts (A), inside-out vesicles (O), and inside-out vesicles reconstituted with purified spectrin dinner (") Actin binding at 25 C and reconstitution of vesicles were performed as described in Methods Gels : (a) insideout vesicles; (b) inside-out vesicles reconstituted with 100 Pg/ml spectrin dimer; and (c) sealed ghosts TABLE ph Dependence of F-Actin Binding to Spectrin-Reconstituted Inside-Out Vesicles PH I Actin binding pglmg vesicle protein Inside-out vesicles were reconstituted with 25 pg/ml spectrin dimer as described in Methods Actin binding was measured at the indicated ph values at an actin concentration of 100,ug/ml Values shown are the means of three independent measurements and have standard deviations of 10-17% Actin binding to reconstituted vesicles depended upon ph, there being, on the average, a 28% decrease in binding for each ph unit above ph 60 (Table 1) Our binding studies were routinely done at ph 65, at which ph the binding was ^-36% greater than at the physiological ph of 7 5 The spectrin dimer, which consists of two polypeptide chains, band 1 (240,000 daltons), and band 2 (220,000 daltons) can self-associate to form a tetramer, (1 + 2)2, which can be separated from the dimer by column chromatography (Fig 2) Fig 3 shows that actin binding to vesicles increased with added spectrin but was independent of whether the spectrin was in the dimeric or tetrameric form Fig 3 also shows that spectrinstimulated actin binding was observed at 25 C but not at 4 C, a finding that correlates well with our observation' that spectrin cross-linking of F-actin in solution is much greater at 25 C than 4 C Because other actin-binding proteins are known to cause F- actin to sediment under conditions similar to the ones used here but in the absence of vesicles (6), we considered the possibility that residual spectrin not bound to vesicles during reconstitution was responsible for the increase in sedimentable F-actin However, spectrin dimer or tetramer alone, at concentrations up to 10 hg/ml, in the absence of vesicles, failed to cause F-actin to sediment under the conditions used (The concentration 10 )ig/ml was chosen because we estimated that after reconstitution with spectrin no more than 5-10% of the supernate could have remained behind, and the highest con- RAPID COMMUNICATIONS 69 5

3 FIGURE 2 Chromatographic separation of spectrin dimers and tetramers, performed as described in Methods Gels : (a) spectrin dimer, ( b) spectrin tetramer The minor bands seen in the gels are proteolytic fragments of spectrin of >120,000 mol wt decreased thereafter because of a failure of the vesicles to sediment at high actin concentrations under the centrifugation conditions used By determining the amount of spectrin on reconstituted vesicles by densitometry of gels, and assuming an average of 1,000 actin monomers per filament, we calculate that at 0 2 mg/ml actin there was one actin filament bound for every 5-10 spectrin molecules on the vesicle It is, of course, possible that single filaments could be anchored by more than one spectrin molecule Whereas spectrin alone can cross-link and bind to actin filaments (see footnote 1 and reference 5), recent studies have shown that erythrocyte membrane protein band 4 1 can stimulate or enhance this interaction (see footnote 1 and references 1 I and 28) Consistent with this idea, Fig 4 shows that, in the presence of spectrin, addition of band 41 doubled the actinbinding capacity of vesicles, whereas in the absence of added spectrin, addition of band 41 had little effect on actin binding It should be noted, however, that the spectrin-depleted vesicles still contain nearly all of their original band 4 1 Inside-out vesicles that are not reconstituted with spectrin retain 5-10% of the spectrin content and, as indicated above, 90% or more of the band 4 1 content of ghost membranes (measured by densitometry of gels) To determine whether these residual proteins were responsible for the binding of F- actin to unreconstituted vesicles, we trypsinized the vesicles and measured their subsequent ability to bind actin Fig 5 shows that trypsinization reduced the actin-binding capacity of vesicles by 70%, whereas SDS gels of these same vesicles (before addition of actin) show that no detectable loss or degradation of spectrin had occurred Since membrane-bound spectrin is known to be trypsin sensitive (21), this lack of degradation could be explained if the spectrin were trapped within the small percentage of sealed right-side-out vesicles that invariably contaminate inside-out vesicles The gels do show that several protein bands were degraded by the trypsin, most notably bands 4 1, 4 2, 3, and 6 Separate FIGURE 3 F-actin binding to spectrin-reconstituted vesicles Vesicles were reconstituted as described in Methods with increasing concentrations of spectrin dimer (") or tetramer (A) Following reconstitution, binding of 80 jig/ml [ 3 H]F-actin was measured at a vesicle concentration of 025 mg/ml at either 4 C (dashed lines) or 25 C (solid lines) Gels : vesicles reconstituted with 0 (a), 10 (b), 25 (c), 50 (d), and 100 (e) jg/ml spectrin tetramer, and with 0 (f), 10 (g), 25 (h), 50 (i), 100 (j) fog/ml spectrin dimer (k) Erythrocyte ghosts At the highest spectrin concentration, 75% of the added F- actin was bound to vesicles centration of spectrin used was 100 f-lg/ml ) In addition, supernates obtained by sedimentation of spectrin-reconstituted vesicles that had been incubated under actin-binding conditions (without added actin) failed to stimulate sedimentation of subsequently added F-actin The data of Fig 3 show that the tetramer and the dimer stimulated actin binding to nearly the same extent, and the gels show that there was an equivalent amount of total spectrin bound in both cases That per unit weight there are half as many tetramers as dimers implies that tetramers are twice as efficient as dimers at binding actin to the membranes To determine how much actin could be bound to reconstituted vesicles, we measured actin binding at actin concentrations ranging from 005 to 1 0 mg/ml Binding increased to a maximum of 315 ttg/mg vesicle protein at 02 mg/ml actin but FIGURE 4 F-actin binding to vesicles with or without added spectrin dimer or band 41 Vesicles were reconstituted with 100 jlg/ml spectrin dimer as described in Methods The reconstituted or unreconstituted vesicles at a protein concentration of 0125 mg/ml were tested for binding of 100 fig/ml F-actin at 25 C as described in Methods with or without the inclusion of 25 pg/ml band 4 1 Gels : (a) purified band 41, ( b) ghosts 696 RAPID COMMUNICATIONS

4 FIGURE 5 Reduction of F-actin binding to unreconstituted vesicles by trypsinization Vesicles at a concentration of 0 5 mg/ml were incubated with the indicated concentrations of trypsin in binding buffer (ph 76) at 25 C for 05 h, at which time a 15-fold weight excess (relative to trypsin) of PMSF was added to all samples (including control) Vesicles were washed once in binding buffer containing 15 pg/ml PMSF and resuspended for measurement of F- actin binding at 25 C at an actin concentration of 80pg/ml Gel : (a) control vesicles, 0pg/ml trypsin; (b) 051ag/ml trypsin ; and (c) 1 0 kg/ml trypsin Sedimentation of vesicles for the binding assay resulted in % vesicle recovery at all trypsin concentrations, measured as described in Methods experiments (not shown) demonstrate that vesicles devoid of band 6 bind as much actin as those with band 6 Further work is in progress to determine which of the remaining proteins may bind actin to the membrane In light of the results described above, however, we suggest that the band 41 protein on the membrane may be responsible for at least part of the observed F-actin binding to unreconstituted vesicles DISCUSSION The results of our experiments show that spectrin can anchor actin filaments to erythrocyte membranes and that band 41 may be importantly involved in this association Because it has been reported that spectrin can bind to band 4 1 (25), our findings imply that spectrin and actin may both associate with band 4 1 on the membrane and share it as a common membrane attachment site Thus, band 41 may play the dual role of helping to anchor the erythrocyte cytoskeleton to the membrane and promoting the association of the two major cytoskeletal components, spectrin and actin Our findings are consistent with the known geometry of the actin- and membrane- binding sites on the spectrin molecule Spectrin binds to inside-out vesicles through a high-affinity association with the membrane protein band 2 1 (3, 12, 29) The work of Tyler et al (25) shows that the 2 1 binding site is near the head of the dimer or near the middle of the tetramer, which is formed by head-to-head association of two dimers (19, 27) More recently work, using the technique of low-angle shadowing,' demonstrates that spectrin's actin-binding sites lie at the tails of the tetramer molecule, making each tetramer 2 Cohen, C M, J M Tyler, and D Branton Manuscript submitted for publication bivalent in actin-binding sites and consequently able to crosslink adjacent actin filaments Thus, dimers can bind to membranes at a site near their head and would have one actinbinding site available at their free tail, whereas tetramers could bind to membranes near their middle and would have two actin-binding sites available Whereas band 41 may be required for spectrin-stimulated actin binding, we cannot conclude from our findings whether band 4 1 is specifically involved in actin binding to unreconstituted vesicles The reduction in actin binding to vesicles by proteolysis shown in Fig 5 could simply reflect reduction of nonspecific (possibly electrostatic) binding caused by loss of vesicle protein rather than loss of a specific protein Our current findings are in contrast to but consistent with our previous report, which showed that purified spectrin dimers would not stimulate binding of added G-actin to vesicles, whereas an oligomeric complex of spectrin, actin, and band 4 1 would (8) In that report we put forth a proposal, subsequently partially confirmed by Brenner and Kom (4), that the oligomeric complex stimulated binding because it itself bound to the membrane and contained actin filament seeds that stimulated polymerization and growth of F-actin from the membrane Because those experiments (8) were performed at 4 C and low G-actin concentration (40 Fig/ml), filament formation without added nucleation sites would have been slow Consequently, in the absence of the complex, there would have been few filaments to bind to the reconstituted vesicles Furthermore, our current findings show that there is no stimulation of F- actin binding by spectrin dimers at 4 C, the temperature at which binding was measured previously Spectrin's similarity to filamin and actin-binding protein from macrophages has recently been demonstrated structurally by low-angle shadowing (26) and biochemically by its ability to cross-link and gel actin filaments (see footnote 1 and references 5, 9, 11, 28) It seems likely that these similarities would arise from a common intracellular function Although the binding of filamin or actin-binding protein or plasma membranes has yet to be demonstrated, the results of our experiments suggest that an important role of such proteins, with or without such accessory proteins as band 41, may be to anchor actin filaments to plasma membranes We thank Catherine Korsgren for her excellent technical assistance in the preparation of actin for these studies This work was supported by grants from the U S Public Health Service (HL24382) and the American Heart Association (research grant-in-aid ), central Massachusetts Division Receivedfor publication 7 April 1980, and in revisedform 12 May 1980 Note Added in Proof After this work was completed, similar conclusions were reached by V Fowler, E J Luna, W R Hargreaves, D L Taylor, and D Branton (manuscript, submitted for publication) using different techniques REFERENCES 1 Ash, J F, D Louvard, and S J Singer 1977 Antibod y induced linkages of plasma membrane proteins to intracellular actomyosin-containing filaments in cultured fibroblasts Proc Natl Acad Set USA 74: Avruch, J, and G Fairbanks 1974 Phosphorylation ofendogenous substrates by erythrocyte membrane protein kinases 1 A monovalent cation-stimulated reaction Biochemistry 13 : Bennett, V, and P J Stenbuck 1979 Identification and partial purification of ankyrin, the high affinity membrane attachment site for humanerythrocyte spectrin J Biol Chem 254: Brenner, S L, and E D Korn 1980 Spectrin/acti n complex isolated from sheep erythrocytes accelerates actin polymerization by simple nucleation J Biol Chem 255 : Brenner, S L, and E D Kom 1979 Spectrin-actin interaction Phosphorylated and RAPID COMMUNICATIONS 697

5 dephosphorylated spectrin tetramer crosslink F-actin J Biol Chem 254: Brotschi, E A, J H Hartwig, and T P Stossei 1978 The gelation of actin by actinbinding protein J Biol Chem 254 : Cohen, C M, P L Jackson, and D Branton 1978 Actin-membrane interactions : association ofg-actin with the red cell membrane J Supramol Struct 9 : Cohen, C M, and D Branton, 1979 The role of spectrin in erythrocyte membrane I stimulated actin polymerization Nature (Land) 279 : Cohen, C M 1979 Lack of correlation between actin polymerization and gelatin induced by human spectrin J Cell Biol 83(2, Pt 2) :308 a (Abstr) 10 Fairbanks, G, T L Steck, and D F H Wallach 1971 Electrophoretic analysis of the major polypeptides of the human erythrocyte membrane Biochemistry 10: It Fowler, V, and D L Taylor 1980 Spectrin plus band 4 1 cross-link actin : regulation by micromolar calcium J Cell Biol 85 : Luna, E, J, G H Kidd, and D Branton 1979 Identification by peptide analysis of the specirin-binding protein in human erythrocytes J Biol Chem, 254: Mooseker, M S, and L G Tilney 1975 Organizatio n of an actin filament-membrane complex Filament polarity and membrane attachment in the microvilli of intestinal epithelial cells J Cell Biol 67 : Pollard, T D,and E D Kom 1973 Electronmicroscopic identification of actin associated with isolated amoeba plasma membrane J Biol Chem 248 : Pollard, T D, and R R Weihing 1974 Actin, myosin and cell movement CRC Crit Rev Biochem, 2: Ralston, G B, J Dunbar, and M White 1977 The temperature dependent dissociation of spectrin Biochem Biophys Acia 491 : Sheets M P, R, G Painter, and S J Singer 1976 Relationships of the spectrin complex of human erythrocyte membranes to the actomyosins of muscle cells Biochemistry 15: Sheetz, M P 1979 DNase I-dependent dissociation of erythrocyte cytoskeletons J Cell Biol 81 : Shotton, D M, B E Burke, and D Branton 1979 The molecular structure of human erythrocyte spectrin Biophysical and electron microscopic studies J Mal Biol 131 : Spudich, J A, and S J Watt 1971 The regulation ofrabbit skeletal muscle contraction 1 Biochemical studiesof the interaction ofthe tropomyosin-troponin complex with actin and the proteolytic fragments of myosin J Biol Chem 246 : Steck, T L 1972 The organization of proteins in human erythrocyte membranes In Membrane Research C F Fox editor Academic Press, Inc New York Steck T L 1974 The organization ofproteins in the human red blood cell membrane J Cell Brol 62 : Steck, T L, and 1 A Kant 1974 Preparation of impermeable ghosts and inside-out vesicles from human erythrocyte membranes Methods Enzymol 31 : Tilney, L G, and P Detmers 1975 Actin in erythrocyte ghosts and its association with spectrin: evidence for a nonfilamentous form of these two molecules in situ J Cell Biol 66: Tyler, J M,WR Hargreaves, and D Branton 1979 Purification of two spectrin binding proteins : biochemical and electron microscopic evidence for site specific reassociation between spectrin and bands 21 and 4 1 Proc Nail Acad Sci U S A 76 : Tyler, J M, J M Anderson, and D Branton 1980 Structural comparison of several actin-binding macromolecules J Cell Biol 85: Ungewickell, E, and W Gratzer 1978 Self association of human spectrin : a thermodynamic and kinetic study Eur J Biochem 88: Ungewickell, E, P M Bennett, R Calved, V Ohanian, and W B Gratzer 1979 In vitro formation of a complex between cytoskeletal proteins of the human erythrocyte Nature (Loud) 280 : Yu, J, and S R Goodman 1979 Syndeins: the spectrin-binding protein(s) ofthe human erythrocyte membrane Proc Nail Acad Sci U S A 76 : , 698 RAPID COMMUNICATIONS